ci: build GreptimeDB binary for later use (#3244)

* ci: build GreptimeDB binaries for later use

* debug CI

* try larger runner host

* Revert "try larger runner host"

This reverts commit 03c18c0f51.

* fix: resolve PR comments

* revert some unrelated action yamls

* fix CI

* use artifact upload v4 for faster upload and download speed

.github/actions/upload-artifacts/action.yml

@@ -6,7 +6,7 @@ inputs:
     required: true
   target-file:
     description: The path of the target artifact
-    required: true
+    required: false
   version:
     description: Version of the artifact
     required: true
@@ -18,6 +18,7 @@ runs:
   using: composite
   steps:
     - name: Create artifacts directory
+      if: ${{ inputs.target-file != '' }}
       working-directory: ${{ inputs.working-dir }}
       shell: bash
       run: |
@@ -51,13 +52,13 @@ runs:
     # Note: The artifacts will be double zip compressed(related issue: https://github.com/actions/upload-artifact/issues/39).
     # However, when we use 'actions/download-artifact@v3' to download the artifacts, it will be automatically unzipped.
     - name: Upload artifacts
-      uses: actions/upload-artifact@v3
+      uses: actions/upload-artifact@v4
       with:
         name: ${{ inputs.artifacts-dir }}
         path: ${{ inputs.working-dir }}/${{ inputs.artifacts-dir }}.tar.gz
 
     - name: Upload checksum
-      uses: actions/upload-artifact@v3
+      uses: actions/upload-artifact@v4
       with:
         name: ${{ inputs.artifacts-dir }}.sha256sum
         path: ${{ inputs.working-dir }}/${{ inputs.artifacts-dir }}.sha256sum

.github/workflows/develop.yml

@@ -78,8 +78,8 @@ jobs:
       - name: Run taplo
         run: taplo format --check
 
-  sqlness:
-    name: Sqlness Test
+  build:
+    name: Build GreptimeDB binaries
     if: github.event.pull_request.draft == false
     runs-on: ${{ matrix.os }}
     strategy:
@@ -89,15 +89,47 @@ jobs:
     steps:
       - uses: actions/checkout@v3
       - uses: arduino/setup-protoc@v1
-        with:
-          repo-token: ${{ secrets.GITHUB_TOKEN }}
       - uses: dtolnay/rust-toolchain@master
         with:
           toolchain: ${{ env.RUST_TOOLCHAIN }}
-      - name: Rust Cache
-        uses: Swatinem/rust-cache@v2
+      - uses: Swatinem/rust-cache@v2
+      - name: Build greptime binaries
+        shell: bash
+        run: cargo build
+      - name: Pack greptime binaries
+        shell: bash
+        run: |
+          mkdir bins && \
+          mv ./target/debug/greptime bins && \
+          mv ./target/debug/sqlness-runner bins
+      - name: Print greptime binaries info
+        run: ls -lh bins
+      - name: Upload artifacts
+        uses: ./.github/actions/upload-artifacts
+        with:
+          artifacts-dir: bins
+          version: current
+
+  sqlness:
+    name: Sqlness Test
+    if: github.event.pull_request.draft == false
+    needs: build
+    runs-on: ${{ matrix.os }}
+    strategy:
+      matrix:
+        os: [ ubuntu-20.04 ]
+    timeout-minutes: 60
+    steps:
+      - uses: actions/checkout@v3
+      - name: Download pre-built binaries
+        uses: actions/download-artifact@v4
+        with:
+          name: bins
+          path: .
+      - name: Unzip binaries
+        run: tar -xvf ./bins.tar.gz
       - name: Run sqlness
-        run: cargo sqlness
+        run: ./bins/sqlness-runner -c ./tests/cases --bins-dir ./bins
       - name: Upload sqlness logs
         if: always()
         uses: actions/upload-artifact@v3
@@ -109,6 +141,7 @@ jobs:
   sqlness-kafka-wal:
     name: Sqlness Test with Kafka Wal
     if: github.event.pull_request.draft == false
+    needs: build
     runs-on: ${{ matrix.os }}
     strategy:
       matrix:
@@ -116,19 +149,18 @@ jobs:
     timeout-minutes: 60
     steps:
       - uses: actions/checkout@v3
-      - uses: arduino/setup-protoc@v1
+      - name: Download pre-built binaries
+        uses: actions/download-artifact@v4
         with:
-          repo-token: ${{ secrets.GITHUB_TOKEN }}
-      - uses: dtolnay/rust-toolchain@master
-        with:
-          toolchain: ${{ env.RUST_TOOLCHAIN }}
-      - name: Rust Cache
-        uses: Swatinem/rust-cache@v2
+          name: bins
+          path: .
+      - name: Unzip binaries
+        run: tar -xvf ./bins.tar.gz
       - name: Setup kafka server
         working-directory: tests-integration/fixtures/kafka
         run: docker compose -f docker-compose-standalone.yml up -d --wait
       - name: Run sqlness
-        run: cargo sqlness -w kafka -k 127.0.0.1:9092
+        run: ./bins/sqlness-runner -w kafka -k 127.0.0.1:9092 -c ./tests/cases --bins-dir ./bins
       - name: Upload sqlness logs
         if: always()
         uses: actions/upload-artifact@v3

docker/dev-builder/ubuntu/Dockerfile

@@ -40,7 +40,7 @@ RUN apt-get -y purge python3.8 && \
 # wildcard here. However, that requires the git's config files and the submodules all owned by the very same user.
 # It's troublesome to do this since the dev build runs in Docker, which is under user "root"; while outside the Docker,
 # it can be a different user that have prepared the submodules.
-RUN git config --global --add safe.directory ‘*’
+RUN git config --global --add safe.directory *
 
 # Install Python dependencies.
 COPY $DOCKER_BUILD_ROOT/docker/python/requirements.txt /etc/greptime/requirements.txt

tests/runner/src/env.rs

@@ -57,6 +57,11 @@ pub struct Env {
     data_home: PathBuf,
     server_addr: Option<String>,
     wal: WalConfig,
+
+    /// The path to the directory that contains the pre-built GreptimeDB binary.
+    /// When running in CI, this is expected to be set.
+    /// If not set, this runner will build the GreptimeDB binary itself when needed, and set this field by then.
+    bins_dir: Arc<Mutex<Option<PathBuf>>>,
 }
 
 #[async_trait]
@@ -78,11 +83,17 @@ impl EnvController for Env {
 }
 
 impl Env {
-    pub fn new(data_home: PathBuf, server_addr: Option<String>, wal: WalConfig) -> Self {
+    pub fn new(
+        data_home: PathBuf,
+        server_addr: Option<String>,
+        wal: WalConfig,
+        bins_dir: Option<PathBuf>,
+    ) -> Self {
         Self {
             data_home,
             server_addr,
             wal,
+            bins_dir: Arc::new(Mutex::new(bins_dir)),
         }
     }
 
@@ -90,7 +101,7 @@ impl Env {
         if let Some(server_addr) = self.server_addr.clone() {
             self.connect_db(&server_addr)
         } else {
-            Self::build_db().await;
+            self.build_db();
             self.setup_wal();
 
             let db_ctx = GreptimeDBContext::new(self.wal.clone());
@@ -116,7 +127,7 @@ impl Env {
         if let Some(server_addr) = self.server_addr.clone() {
             self.connect_db(&server_addr)
         } else {
-            Self::build_db().await;
+            self.build_db();
             self.setup_wal();
 
             let db_ctx = GreptimeDBContext::new(self.wal.clone());
@@ -249,8 +260,12 @@ impl Env {
         #[cfg(windows)]
         let program = "greptime.exe";
 
+        let bins_dir = self.bins_dir.lock().unwrap().clone().expect(
+            "GreptimeDB binary is not available. Please pass in the path to the directory that contains the pre-built GreptimeDB binary. Or you may call `self.build_db()` beforehand.",
+        );
+
         let mut process = Command::new(program)
-            .current_dir(util::get_binary_dir("debug"))
+            .current_dir(bins_dir)
             .env("TZ", "UTC")
             .args(args)
             .stdout(log_file)
@@ -374,7 +389,11 @@ impl Env {
     }
 
     /// Build the DB with `cargo build --bin greptime`
-    async fn build_db() {
+    fn build_db(&self) {
+        if self.bins_dir.lock().unwrap().is_some() {
+            return;
+        }
+
         println!("Going to build the DB...");
         let output = Command::new("cargo")
             .current_dir(util::get_workspace_root())
@@ -389,7 +408,12 @@ impl Env {
             io::stderr().write_all(&output.stderr).unwrap();
             panic!();
         }
-        println!("Build finished, starting...");
+
+        let _ = self
+            .bins_dir
+            .lock()
+            .unwrap()
+            .insert(util::get_binary_dir("debug"));
     }
 }
 

tests/runner/src/main.rs

@@ -62,6 +62,11 @@ struct Args {
     /// from starting a kafka cluster, and use the given endpoint as kafka backend.
     #[clap(short, long)]
     kafka_wal_broker_endpoints: Option<String>,
+
+    /// The path to the directory where GreptimeDB's binaries resides.
+    /// If not set, sqlness will build GreptimeDB on the fly.
+    #[clap(long)]
+    bins_dir: Option<PathBuf>,
 }
 
 #[tokio::main]
@@ -94,6 +99,9 @@ async fn main() {
         },
     };
 
-    let runner = Runner::new(config, Env::new(data_home, args.server_addr, wal));
+    let runner = Runner::new(
+        config,
+        Env::new(data_home, args.server_addr, wal, args.bins_dir),
+    );
     runner.run().await.unwrap();
 }

tests/runner/src/util.rs

@@ -91,7 +91,7 @@ pub fn get_workspace_root() -> String {
     runner_crate_path.into_os_string().into_string().unwrap()
 }
 
-pub fn get_binary_dir(mode: &str) -> String {
+pub fn get_binary_dir(mode: &str) -> PathBuf {
     // first go to the workspace root.
     let mut workspace_root = PathBuf::from(get_workspace_root());
 
@@ -99,7 +99,7 @@ pub fn get_binary_dir(mode: &str) -> String {
     workspace_root.push("target");
     workspace_root.push(mode);
 
-    workspace_root.into_os_string().into_string().unwrap()
+    workspace_root
 }
 
 /// Spin-waiting a socket address is available, or timeout.